read

MEDIUM W011: Third-party content exposure detected (indirect prompt injection risk).

• Third-party content exposure detected (high risk: 0.90). This skill explicitly fetches and converts arbitrary public URLs (including X/Twitter, WeChat, GitHub, and other web pages) via the proxy cascade and local fetch methods described in SKILL.md and references/read-methods.md, causing the agent to ingest untrusted, user-generated third-party content as part of its required runtime workflow.

MEDIUM W012: Unverifiable external dependency detected (runtime URL that controls agent).

• Potentially malicious external URL detected (high risk: 0.80). The fetch.sh script calls "npx --yes agent-fetch" at runtime (which pulls and executes code from the npm registry, e.g. registry.npmjs.org), so remote code is fetched and executed as a runtime fallback dependency.